I'm guessing here, but it looks like you're loading the URDF into the
parameter called urdf_xml. You should load it into robot_description,
and it should work as usual.

I forget where I borrowed this convention from, but the idea was to
allow you to specify which parameter to read the urdf from. So if you
normally load the URDF into hand_description, then you could set
urdf_xml to hand_description, and then it would load hand_description
instead of robot_description.

-David!!

On Thu, Sep 30, 2010 at 9:40 AM, Ugo Cupcic <ugo@shadowrobot.com> wrote:
>  Hey,
>
> Thanks for the useful tools!
>
> Is there any requirements concerning the urdf to provide the
> arm_kinematics package?
>
> I'm trying to use it for our arm and every time I'm trying to load the
> model, it just tells me:
> [FATAL] [1285854990.845924280]: Could not load the xml from parameter
> server:
> --- and prints the full urdf model here.
> ---
>
> Changing the beginning of the init() method like this solved the
> problem. Any ideas?
>
> bool Kinematics::init() {
>    // Get URDF XML
>    std::string urdf_xml, full_urdf_xml;
>
>    std::string result;
>    nh.param("urdf_xml",result,std::string("robot_description"));
>    /*
>    nh.searchParam(urdf_xml,full_urdf_xml);
>    ROS_DEBUG("Reading xml file from parameter server");
>    std::string result;
>    if (!nh.getParam(full_urdf_xml, result)) {
>        ROS_FATAL("Could not load the xml from parameter server: %s",
> urdf_xml.c_str());
>        return false;
>    }
>    */
>
>
> Cheers,
>
> Ugo
>
> On 29/09/10 20:02, Wim Meeussen wrote:
>>> Thank you for creating these tools!  I am sure they will come in handy in
>>> lots of situations.
>>>
>>> Might I suggest renaming joint_controller to joint_state_publisher or
>>> fake_joint_state_publisher?  Also, is easy_state_publisher is a special case
>>> of joint_controller?
>> I really like the idea of the joint_state_publisher gui, it makes
>> visualizing and debugging a urdf file much easier. This would even
>> allow you to use rviz to debug a urdf file. This is something we
>> should absorb into the robot_model stack (we'll have to look into the
>> dependencies).
>>
>> For the easy_state_publisher, I'd think that if we give the new
>> joint_state_publisher the ability to run without a gui, it would be
>> possible to run it in combination with the current
>> robot_state_publisher offers the same (and much more!) functionality.
>> What do you think?
>>
>> Wim
>>
>>
>
> --
> Ugo Cupcic         |  Shadow Robot Company | ugo@shadowrobot.com
> Software Engineer  |    251 Liverpool Road |
> need a Hand?       |    London  N1 1LX     | +44 20 7700 2487
> http://www.shadowrobot.com/hand/              @shadowrobot
>
> _______________________________________________
> ros-users mailing list
> ros-users@code.ros.org
> https://code.ros.org/mailman/listinfo/ros-users
>